Hybrid Laser Printing of 3D, Multiscale, Multimaterial Hydrogel Structures

Abstract

Fabrication of multiscale, multi-material three-dimensional (3D) structures at high resolution is difficult using current technologies. This is especially significant when working with hydrated and mechanically weak hydrogel materials. In this work, a new hybrid laser printing (HLP) technology is reported to print complex, multiscale, multimaterial, 3D hydrogel structures with microscale resolution. This technique of fabrication utilizes sequential additive and subtractive modes of material fabrication, that are typically considered as mutually exclusive due to differences in their material processing conditions. Further, compared to current laser writing systems that enforce stringent processing depth limits, HLP is shown to fabricate structures at any depth inside the material. As a proof-of-principle, a Mayan Pyramid with embedded cube-frame is printed using model synthetic polyethylene glycol diacrylate (PEGDA) hydrogel. Printing of ready-to-use open-well chips with embedded microchannels is also demonstrated using PEGDA and gelatin methacrylate (GelMA) hydrogels for potential applications in biomedical sciences. Next, HLP is used in additive and additive modes to print multiscale 3D structures spanning in size from centimeter to micrometers within minutes, which is followed by printing of 3D, multi-material, multiscale structures using this technology. Overall, this work demonstrates that HLP’s fabrication versatility can potentially offer a unique opportunity for a range of applications in optics and photonics, biomedical sciences, microfluidics, soft robotics, etc.

1. Introduction

Over the course of billions of years, nature has created and refined numerous elegant biosynthetic processes to make sophisticated functional structures. In contrast, current manufacturing techniques are still limited in their ability to fabricate 3D, multiscale, multi-material structures. New advances in fabrication methods are expected to revolutionize the development of next-generation devices for optics, photonics, microfluidics, soft robotics, flexible electronics, organ-on-a-chip and many other applications.^[1,2,3] Already, the emergence of additive manufacturing methods such as extrusion-based fused deposition modeling (FDM), inkjet-based multiJet modeling (MJM), digital light processing (DLP) and stereolithography (SLA), have allowed the fabrication of structures and devices with high design flexibility, in a highly automated, assembly-free, and low cost manner.^[2,4] A new version of DLP, coined as continuous liquid interface production (CLIP), is able to fabricate centimeter-sized polymeric constructs in a quick and continuous fashion.^[5] However, these methods cannot fabricate structures with resolution better than 25 μm. Multiphoton polymerization (MPP), often coined as direct laser writing, is an additive manufacturing method that allows the printing of complicated 3D structures at sub-micrometer resolution.^[6,7] However, slow point-by-point scanning approach generally makes this method impractical for printing centimeter scale multiscale structures.

As compared to most of the additive methods discussed above, subtractive manufacturing methods based on lithography provide similar resolution for fabricating micro- and nano-scale structures, but these methods are typically limited to planar or 2.5D structures.^[8] Multiphoton ablation (MPA), a method that utilize ultrafast lasers to remove material, is capable of creating embedded hollow micro-features such as channels, grooves in glass and polymer substrates.^[9,10] Similar to MPP, MPA also suffers from low scalability and throughput due to the serial nature of fabrication. Additionally, the dependence of MPA on the optical transparency of the material, limits its penetration depth and overall design flexibility. Few research groups have utilized the ability of ultrafast lasers to process material in both additive (MPP) and subtractive (MPA) modes to develop a hybrid additive-subtractive method.^[10,11,12] But, current laser based hybrid methods are limited by scalability, types of specialized materials, and incompatible laser processing requirements, thereby preventing its widespread use in the field.

In this work, we report the design and development of a hybrid laser printing (HLP) technology, that combines the key advantages of CLIP process (quick on-demand continuous fabrication) and MPP and MPA processes (high-resolution and superior design flexibility). Using a series of proof-of-principle experiments, we show that HLP is capable of printing 3D, multiscale, multi-material structures using model biocompatible hydrogel materials that are highly difficult and/or extremely time consuming to fabricate using current technologies. We demonstrate the HLP versatility of printing in three different modes of fabrication: additive/subtrative, additive/additive and multimaterial modes and we foresee that HLP’s fabrication capabilites can be applied to a broad range of applications.

2. Design of hybrid laser printer

HLP setup consists of a femtosecond laser source (Coherent, Ti:Sapphire), which is used in both additive and subtractive modes of fabrication. The additive crosslinking mode, indicated by solid blue arrows in Figure 1, is created by passing the fs-laser beam through a second harmonic generator (SHG) to obtain ultraviolet wavelengths and then spatially modulated via a digital micromirror device (DMD). DMD is an electronic board embedded with an array of micromirrors. Based on a user-defined image, DMD can selectively switch mirrors into either an ON state or an OFF state and create a light pattern that selectively crosslinks photosensitive prepolymer into 2D layers of a defined thickness. The subtractive ablation mode, indicated by the dashed red arrows in Figure 1, utilizes fs-laser beam directed via an objective lens to ablate voids within the previously crosslinked layer using MPA. Additionally, fs-laser can be also used in additive mode to polymerize 3D structures with microscale resolution using MPP. In this prototype, the additive mode can print ~1 cm² feature in XY with smallest feature size of 30 μm, while the subtractive MPA mode can fully ablate features with a minimum feature size of 3 μm, based on the absorption properties of the hydrogel prepolymer. Figure 1 inset (big dashed box) shows the one sequence of the process flow of additive DMD crosslinking (CLIP) and subtractive MPA. Figure 1 inset (small dashed box) shows the schematic diagram of sample holder, stage and Teflon window. A detail description of the optical setup of HLP and a photograph and detail description of the sample holder and elevating stage is described in the Methods section and Supp. Info (Figure S1).

Figure 1.

(A) Schematic of femtosecond laser-based additive-subtractive/additive-additive HLP machine. [FS-femtosecond laser, SHG-second harmonic generator, I-isolator, BS-beam splitter, P-polarizer, HWP-half waveplate, SHT-shutter, L-lens, PH-pin hole, FM-mirror in a flip mount, D-diffuser, DMD-digital micromirror device, DM-dichroic mirror, CM-camera, OL-objective, S-sample, SH-sample holder, ST-stage, TF-Teflon film]. Small dashed box: Schematic showing sample holder (SH) with teflon film (TF) and elevating stage (ST). Enlarged dashed box: Process flow of HLP showing one sequence of additive crosslinking and subtractive ablation steps.

3. Additive/subtractive mode of Hybrid laser printer

We characterized the dead-zone and ablation z-range of Hybrid laser printer. Fundamental to the HLP process is the relationship between the ‘dead-zone’ from the additive CLIP mode, and ‘ablation z-range’ from subtractive MPA mode (Figure 2A). Characterization of dead-zone is necessary for accurate determination of speed and z-resolution during the additive step while assessment of ablation z-range is important to determine the exact location of laser focus during the subtractive mode. Dead-zone is a thin uncrosslinked prepolymer solution between the Teflon window and the crosslinked hydrogel structure attached to the stage. This zone is created due to the inhibition of the photo-crosslinking process, a result of diffusion of oxygen through the permeable Teflon window. This prevents the adhesion of newly crosslinked layers to the bottom window and the liquid hydrogel prepolymer solution can freely refill the fabrication area adjacent to the Teflon window. The dead-zone thickness can vary in the order of tens of micrometers depending upon the control parameter such as laser intensity, exposure time, crosslinking properties of hydrogel, and local oxygen concentrations. Furthermore, the dead-zone defines the relationship between print speed and part resolution in the additive mode, and influences the linewidth of ablation part and penetration depths in the subtractive fabrication mode.

Figure 2.

(A) Schematic showing key parameters in HLP. Ablation z-range and dead-zone are two key parameters, which are highlighted in the figure by curly bracket. Please note that the additive and subtractive processes are sequential processes, although they are depicted in the same figure for simplicity. (B) Plot of dead-zone thickness as the function of exposure time and laser power. As for an instance, PEGDA slab crosslinked using crosslinking laser power of (P_add) of 150 mW and exposure time (t_add) of 20 s (marked by black arrow) results in dead-zone of 30 μm. (C) Schematic cartoon showing lines ablated at different penetration depths and plot of ablated linewidth as a function of laser penetration depths and laser powers. Results shows a decreasing linewidth trend with increasing laser penetration depth. Red marks in the plot represent the upper limit of ablation z-range. (D) Plot of ablated linewidth as a function of scanning speeds and laser powers. Plot shows a decreasing linewidth trend with decreasing laser dose. For all of the studies in this figure, 90% PEGDA and 1% LAP was used as material for fabrication.

‘Ablation z-range’ is the region within the crosslinked hydrogel layers where material can be reliably removed in the subtractive ablation mode of HLP (Figure 2A). The two-photon absorption induced ablation process is limited to certain depth due to light absorption and scattering effects within the hydrogel, that defines the upper end of the ‘ablation z range.’ The lower end of the ‘ablation z-range’ is the region just above the dead-zone within the crosslinked hydrogel layer. Operating in this z-range ensures that the prepolymer solution does not inadvertently crosslink during the subtractive ablation step and avoids forming unwanted bubbles.

For this work, PEGDA prepolymer (90% by weight (wt), 700 MW) mixed with water soluble photoinitiator lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP 1% by wt) was chosen as a model synthetic hydrogel prepolymer solution. Wavelengths of 400 nm and 800 nm were chosen for additive crosslinking (λ_add) and subtractive ablation (λ_sub) modes respectively. In order to estimate the dead-zone, the L shaped stage was placed exactly at 200 μm from the surface of the Teflon window. A spatially modulated laser beam (λ_add=400 nm) was used to crosslink a uniform square shaped structures for a range of crosslinking laser power (P_add) and exposure time (t_add). The thickness of the structures was measured using HIROX microscopy and was subtracted from 200 μm to obtain the dead-zone thickness as a function of P_add and t_add. (Figure 2B). For the laser dosage used in this work, the dead-zone thickness ranges from 8±8 μm to 97±3.5 μm and show an increasing trend with decreasing laser dosage. Increase in dead-zone thickness essentially means that a higher speed can be implemented to fabricate structures.

To measure the ablation z-range, width of the ablated lines was characterized at different depths of crosslinked structure (Figure 2C). First PEGDA slab was crosslinked using crosslinking laser power of (P_add) of 150 mW and exposure time (t_add) of 20 s that results in small dead-zone of 30 μm. Next, lines were ablated within the crosslinked structure using the ablation laser powers of (P_sub) of 800 mW, 1000 mW and 1200 mW, and scanning speed (v_sub) of 100 μm/s. The ablated lines were written at the depth of 40 μm to 160 μm from the surface of Teflon window. Results demonstrate a decrease in linewidth with an increase in laser penetration depth, due to light absorption and scattering within the sample for all the laser powers. For P_sub of 800 mW, 1000 mW and 1200 mW, the upper ends of ablated z-range were measured to be 120 μm, 140 μm and 160 μm respectively. The lower ends of ablation-range were obtained from the dead-zone plot (Figure 2B). For a particular additive step with P_add of 150 mW and t_add of 20 s, the dead-zone thickness was measured as 30 μm, while for a particular subtractive step with P_sub of 1200 mW and v_sub of 100 μm/s, the upper end of the range was measured to be 160 μm, giving an ‘ablation z-range’ of ~35–160 μm. In a similar fashion, ablation z-ranges were calculated for a range of laser powers and scanning speeds, and these parameters were used to automate the HLP platform.

Next, the amount of material removal during the subtractive step of HLP was characterized in term of ablation linewidth. To do so, ablation laser powers (P_sub) and scanning speeds (v_sub) were varied and the widths of the ablated lines were measured. First, hydrogel slabs were crosslinked with P_add of 150 mW and t_add of 20 s. Second, laser ablation was carried out at a depth of 100 μm from the Teflon surface, and width of the ablated lines was measured using an optical microscope and the data was recorded by averaging 5 readings. As shown in Figure 2D, results show that the ablated linewidth decreases from 8 μm to 3 μm with an increase in scanning speeds (v_sub) and a decrease in ablation laser powers (P_sub). These data were used to determine the ablation laser power and scanning speed for achieving desired ablation linewidth.

Our result suggests that the minimum feature size of the ablated line is around 3 μm, and this value is higher compared to resolution based on Abbe’s resolution criteria. Considering Abbe’s criteria, resolution is given by d=λ/(2*NA), where λ is wavelength of light and NA is numerical aperture. For two-photon absorption, the resolution improves roughly by a factor of √2. ^[7] For a laser wavelength of 800 nm and numerical aperture of 0.55, the theoretical resolution of our optical setup should be ~514 nm. However, the mechanism of fs ablation of the hydrogel is dictated by the shockwave generation and bubble formation that extend beyond the laser focus spot and increases the feature size of ablated structure. Moreover, we only report the linewidth of fully ablated lines, and do not take into account partial ablation or incomplete ablation events.

Next HLP was used to print complex 3D structures with internal micro-features. To test the achievable design complexity of HLP, a Mayan pyramid, with a hollow cube-frame embedded within a 3D pyramid structure, was fabricated using 90% PEGDA, 1% LAP prepolymer solution. Briefly, a 3D model of a Mayan pyramid is sliced to generate digital masks for the DMD using a custom written algorithm, and additive-subtractive sequential processes were used to print the structure in an automated manner (Figure 3A, B). For the additive steps, laser wavelength (λ_add) of 400 nm and crosslinking laser power (P_add) of 200 mW was used to selectively crosslink PEGDA layers in a continuous fashion onto a methacrylated glass coverslip. For subtractive steps, a laser wavelength (λ_sub) of 800 nm, an ablation laser power (P_sub) of 1200 mW, and a scanning speed (v_sub) of 100 μm/s was used to ablate a hollow cube-frame within the crosslinked layers. (Figure 3B). The additive and subtractive process was iterated in an automated fashion to fabricate a Mayan pyramid (6 mm tall) with an embedded cube-frame (600 μm³). The location of the cube is 2 mm inside the pyramid from the base of the pyramid (Figure 3C). A total fabrication time of 66 minutes was dominated by the time required for subtractive steps (62 minutes) as compared to all the additive steps combined (3.6 minutes) and the time required to transition between steps (0.4 minutes). This work demonstrates HLP’s ability to remove material at any depth within a 3D structures. As compared to current laser based fabrication methods where laser penetration depth is substantially influenced by the optical properties of the prepolymer solution, HLP is less dependent on the optical properties of the prepolymer solution due to the sequential additive and subtractive modes of fabrication that enables the ablation of materials from any depth within a 3D structure.

Figure 3.

(A) Diagram showing slicing of an stl model of Mayan pyramid to create a stack of digital masks. (B) Key steps in the HLP process to print a Mayan pyramid with embedded cube frame. Iteration of additive and subtractive steps are involved to print a complete ‘cube inside pyramid’ structure (C) Side, top and isometric views of ‘cube inside pyramid’ structure in 90% PEGDA, 1% LAP and the image is recorded by an HIROX digital optical microscope. For additive step laser power of 200 mW and laser wavelength of 400 nm was used, while laser power of 1200 mW, scanning speed of 100μm/s and laser wavelength of 800 nm was used for subtractive steps. The height of the pyramid is 6 mm. The cube is 600 μm³, embedded 2 mm from the bottom of the pyramid. (red arrow points to the embedded cube).

To further demonstrate HLP’s printing capabilities, open-well chips were printed using 10% PEGDA 1% LAP (Figure 4). Multilayer 1.5 mm thick chips that consist of two open rectangular wells (1.25 mm × 2.5 mm) connected by either in-plane microchannels (6 μm in diameter) (Figure 4A) or out-of-plane microchannel (8 μm) were printed (Figure 4B). For both these structures, the total printing time was 4.5 minutes. For chips with embedded in-plane microchannels, DMD-modulated laser beam was used to print rectangular wells using a λ_add of 400 nm, P_add of 150 mW and t_add of 20 s followed by ablation of microchannels in straight, zig-zag and square-wave configurations using a P_sub of 1200 mW and v_sub of 100 μm/s. Repeated scans (2 times) are necessary to ensure that channels are not blocked with any debris. Rhodamine B dye, pipetted in one of the wells, was observed to perfuse within the channels under static hydrostatic pressure as shown in Figure 4A. Chips with out-of-plane 3D microchannels were fabricated using P_add of 150 mW, t_add of 20 s in the additive mode, and P_sub of 1300 mW and v_sub of 50 μm/s in the subtractive mode. Sequential additive and subtractive modes in HLP allow the ablation of 3D channel that span across multiple crosslinked layers (Figure 4B).

Figure 4.

Printing user-defined two-well PEGDA chips with interconnecting in-plane and out-of-plane microchannels. (A) In-plane channels: side and top views of channels with zig-zag and square-wave configurations are depicted. Perfusion of Rhodamine B through zig-zag channels is demonstrated. (B) Out-of-plane channels: Side, isometric and top views of 3D channels spanning multiple crosslinked layers are depicted. White dotted box highlights the zig-zag ablated channels. For all of the studies in this figure, 10% PEGDA an d 1% LAP was used for fabricating the structures.

To demonstrate the utility of HLP for potential bioscience applications, 4-well PEGDA chips with embedded microchannels were printed (Figure 5). To assess the utility of such chips in studying cell-cell communication, we choose model osteocyte cell line. In the bone tissue, osteocytes reside within isolated cavities (lacunae) and they communicate with their neighbouring cells by forming dendritic processes within hollow microchannels called canaliculi. To mimic this microenvironment, PEGDA chips, 200 μm thick, with 40 microchannels (diameter 5 μm) were ablated within walls (width=100 μm) that separate adjacent wells. These chips were printed using a P_add of 150 mW, t_add of 20 s, P_sub of 1200 mW, and v_sub of 200 μm/s. For the fabrication of these chips, a total printing time of 3 minutes was required. Mice MLO-Y4 osteocytes, seeded within the chips, remain isolated in their respective wells; however, they extend cell processes through the microchannels to establish direct physical contact with cells seeded in the adjacent well. This is confirmed by the presence of the nucleus on the either side of the wall, while only cell processes labelled by f-actin (green) is present within the channel (Figure 5A).

Figure 5.

(A) Brightfield image of a 4-well PEGDA chips with inter-well microchannels is depicted. Fluorescent images showing actin (green) and nucleus (blue). MLO-Y4 cells seeded in neighboring wells communicate with each other by extending cell processes within the ablated channels. White arrows point the communication channels with a diameter of 5 μm (B) Brightfield image of a 4-well PEGDA chips with inter-well microchannels is depicted. Seeded osteosarcoma Saos-2 cells migrate within the ablated channels. White arrows show the migration channels with a diameter of 7 μm. For all of the studies in this figure, 10% PEGDA and 1% LAP was used for fabricating the open-well chips.

To potentially extend the utility of these chips to cell migration studies, the chip design was modified by increasing the wall thickness to 300 μm. A P_sub of 1400 mW and v_sub of 50 μm/s was used to ablate channels of diameter 7 μm, a size that would facilitate cell migration within the channel (Figure 5B). Model human Saos-2 osteosarcoma cell line, chosen for this work, is able to migrate within the channels, as indicated by the presence of both f-actin (green) and nucleus (blue) within the interconnecting channels on day 4. For these chips, a total printing time of 11 minutes was required to account for the increased wall thickness and channel sizes. Based on Figure 5A and Figure 5B, we can say that the resulting channels are open and are not clogged by debris. The process is not toxic to cells as seen in the figures, where seeded cells reliably migrate and form cell-cell connections within the microchannels. To demonstrate that HLP can be extended to other photosensitive hydrogels, 4-well chips using naturally-derived gelatin methacrylate (GelMA) hydrogel were printed using a P_add of 200 mW, t_add of 25 s, P_sub of 1000 mW, and v_sub of 100 μm/s. (Figure S2, Supp. Info)

4. Additive/additive mode of Hybrid laser printer

HLP in additive DMD based printing (CLIP) and additive MPP was used to achieve on-demand fabrication of multiscale 3D structures with superior design flexibility. To demonstrate this capability, we choose to use model prepolymer solution (90% PEGDA, 1% LAP) to print a multi-tier design that consists of three log-pile structures printed on different Z-heights and XY locations. Figure 6A illustrates the sequence of steps, while Figure 6B shows the series of masks for additive CLIP and laser paths during the additive MPP steps. In the first step, a 200 μm thick layer was additively printed using CLIP (Mask 1). In the second step, MPP was used to print the first woodpile structure. This is followed by Step 3, where another 200 μm thick layer was printed using Mask 2 (CLIP). Similarly, sequential use of MPP and CLIP was used to print the rest of the structure (Steps 3–6). To clearly visualize the embedded 3D logpile structures in different z-planes without mechanical sectioning, a rhombus shaped blank structure was added to the DMD masks. Figure 6C depicts a representative HIROX digital microscopy image of HLP-printed 3D multiscale structure. In this study, additive CLIP steps used a laser power of 200 mW and exposure time of 5 sec, while additive MPP steps used a laser power of 400 mW and scanning speed of 100 μm/s using an objective lens (Zeiss, 20X) with numerical aperture of 0.25. The logpile structures consists of 10 layers of logpile structure (log-size=2 μm; log-spacing=20 μm). Each logpile structure is ~80 μm thick and each structure is printed at different z-levels/heights. (Figure S3, Supp. Info) The total fabrication time of 45 minutes is mostly dictated by the MPP steps. This work demonstrates that the HLP is capable of fabricating 3D multiscale structure using multistep and automated additive-additive approach.

Figure 6.

HLP’s capability of additive-additive multiscale printing (A) Schematic depicting sequential additive steps of DMD based printing (CLIP) and multiphoton polymerization (MPP). Three woodpile structures were printed in different depth using MPP. (B) Corresponding DMD masks and wood pile structures. (C) HIROX digital optical microscopy image depicting three woodpile structures fabricated at different z-planes and xy spatial locations. Three different woodpile structure marked by 1, 2 and 3 correspond to the zoomed images to the left. Structure was printed using 10% PEGDA and 1% LAP with laser power of 200 mW and exposure time of 5 seconds in additive CLIP step. While high resolution MPP fabrication was performed using laser power of 400 mW and scanning speed of 100 μm/s.

5. Multi-material printing mode of Hybrid laser printer

To demonstrate the HLP’s capability of printing multi-material 3D structures, we built a new fluid chamber with necessary tube connections to enable efficient switching of different prepolymer solutions. The fluid chamber was designed using Autodesk Inventor and machined using a vertical mill. The sample holder consists of the five 3-mm diameter inlets, printing region (also known as fabrication window) of 1 cm in diameter, and five 3-mm diameter outlets. Three inlets were used to pump different hydrogel prepolymer solutions, while the fourth and the fifth inlets were designed to pump in washing solution and/or nitrogen. Similarly, three of outlets were designed to pump-out the prepolymer solution to the respective syringes so that the solutions can be reused. The other two outlets were used to drain the DI water during washing steps; this step is necessary to avoid mixing of different prepolymer solutions during printing. A circular hole (diameter 10 mm) was drilled in the center of the sample holder and a thin Teflon film was glued. This area is named as fabrication window where the exposure of the hydrogel prepolymer solution takes place. The floor of the fluid chamber was sloped so that the prepolymer solutions can be easily drained during the washing steps. Before introducing new prepolymer solution, compressed nitrogen was used to ensure the complete removal of DI water. While printing GelMA structures, the heater (shown in Figure S1, Supp. Info.) was maintained at 40 °C. This setup allowed for repeatable printing of multi-materials structures with no mixing of the prepolymer solutions.

HLP in DMD based additive mode (CLIP) was used to demonstrate the multi-material printing capability in both the z (Mayan Pyramid, Figure 8A) and the xy (Yin-yang, Figure 8B) directions. First, digital masks and image of a Mayan Pyramid printed using three solutions composed of 90% PEGDA hydrogel, 1% LAP photoinitiator and food color dye of different colors (blue, green, and red). This structure was printed using the crosslinking laser power (P_add) of 200 mW, a printing speed in the vertical direction of 0.02 mm/s, and an exposure time per layer (t_add) of 3 seconds. Similarly, yin-yang structure was printed using a laser power of (P_add) 200 mW and exposure time (t_add) of 5 seconds for each step. Second, additive DMD based mode (CLIP) was combined with additive MPP to print nested logpile multiscale 3D structures using three different polymer compositions. (Figure 8C) First, CLIP was used to print the outer thick pink wall using 20% GELMA, 0.25% LAP doped with Rhodamine B dye. Second, CLIP was used to print internal walls (that separate the wells) using 50% PEGDA, 1% LAP. For both these structures, a crosslinking laser power (P_add) of 200 mW and exposure time (t_add) of 5 seconds was used. Third, MPP was used to print high resolution logpile structure using a laser power of 400 mW and scanning speed of 100 μm/s. The laser beam was focused using 20X objective lens (Zeiss, NA=0.25) to obtain a minimum feature size of 2 μm. Next, multi-material 3D printing capability was demonstrated by combining additive CLIP and subtractive MPA modes. Additive CLIP was used to print structures using two hydrogel compositions (90% PEGDA, 0.5% LAP with green food dye, 50% PEGDA, 0.5% LAP, with red food dye) with a laser power (P_add) of 200 mW and exposure time (t_add) of 5 s. (Figure 8D) Laser processing conditions to achieve a specific ablation size within both materials (Figure S4, Supp. Info) were used to ablate an array of channels in both regions using MPP. For this step, laser power was kept constant (P_sub=1500 mW) and the scanning speed was switched between 50 μm/s to 150 μm/s. The plot shown in Figure S4 can be explicitly used to ablate desired linewidth structures in these material. For instance, to print a uniform channel of (~7.6 μm) throughout the both slabs, the scanning speed is reuired to maintain at 130 μm/s inside 90% PEGDA slab and 50 μm/s inside the 50% PEGDA slab.

Figure 8.

(A) Schematic depicting digital mask and multicolored Mayan pyramid fabricated using multimaterial additive printing of three colored 90% PEGDA, 1% LAP prepolymer solutions. (B) Ying and yang structure printed using four different colored 90% PEGDA, 1% LAP. (C) Multiscale printing of multimaterial structure using two commonly used hydrogel (GelMA and PEGDA). The outer thick wall colored in pink is printed with 15% GELMA, 0.5% LAP using DMD based fabrication. Similarly, the DMD based approach is used to print intersecting lines using 50% PEGDA, 1% LAP. Further logpile structure shown inside the black square box is printed using MPP based DLW approach using 90% PEGDA, 1% LAP. (D) Additive and subtractive multimaterial approach of fabrication is demonstrated using 90% PEGDA, 0.5% LAP (colored in green) and 50% PEGDA, 0.5% LAP (colored in red). Insets shows the zoomed version of ablated lines using MPA.

6. Discussion

6.1. Comparison of HLP with existing fabrication techniques

Contrary to materials found in nature that possess 3D structural hierarchy and material heterogeneity, man-made materials remain relatively simple. Current manufacturing technologies are limited by a trade-off between the use of multiple materials, overall size range, dimensionality, throughput and resolution. (Table S1) For instance, subtractive methods based on lithography (photo-, soft-, nanoimprint-lithography) exhibit excellent feature resolution, however these methods typically generate planar devices or they required complicated multiple bonding and stacking steps to fabricate devices with even simple 3D designs.^[8,13] On the other hand, additive manufacturing methods such as FDM, MJM, and DLP offer 3D design flexibility, however achieving microscale resolution with these methods remain challenging.^[2–4] Among the various fabrication methods at our disposal, ultrafast lasers, with their unique property of nonlinear multiphoton absorption, have revolutionized the processing of materials at micrometer scale using MPP and MPA.^[14,15] In this work, we have designed and built a single versatile manufacturing platform coined as HLP. HLP, by combining additive CLIP with additive MPP and subtractive MPA processes, enables quick printing of centimeter-sized hydrogel chips with embedded hollow or solid micro-features; this would otherwise require, multiple planar fabrication followed by the complex alignment of multiple components using conventional lithography. Integration of subtractive MPA with CLIP also ensures reliable removal of material in defined locations. This is a clear advantage as compared to current 3D printing methods where removal of support or sacrificial material from micro-channels/features remains challenging. Using several proof-of-concept studies, we demonstrate HLP’s ability to shape soft hydrated and difficult-to-process hydrogel materials into complex multiscale structures that are either highly challenging, or time consuming to fabricate, or cannot be fabricated using current methods.^[16]

6.2. Comparison of HLP with existing laser based hybrid methods

Only a handful researchers have developed hybrid additive-subtractive methods using lasers due to the material incompatibilities and/or processing requirements.^[10,11] One approach, that has been widely used, has combined ultrafast lasers in additive (MPP) and subtractive (MPA) modes to fabricate complex 3D structures using epoxy based photoresists. However, time-consuming serial nature of the both MPP and MPA has limited the scalability of these methods and therefore their utility in the field for making centimeter sized devices.^[8,12] In comparison, HLP combines additive CLIP with MPP/MPA processes to enable printing of centimeter-sized hydrogel chips with embedded micro-features within minutes. Another laser hybrid method utilizes specialized materials such as Foturan photosensitive glass to make hollow micro-features. This method involves two steps. In step 1, laser irradiation is used to modify the material properties to allow facile removal of materials using chemical etching. In step 2, MPP is used to crosslink complex 3D structure within the chemically etched channels. Since processing requirements for the etching and MPP steps are distinct, this approach cannot be automated into a multi-step multi-layer process. Additionally, the use of harsh chemical and processes (etching and high temperature treatment) makes this process incompatible with hydrogel materials. In comparison, HLP technique can print 3D structures with embedded features at a resolution of few micrometers using a multi-step multi-layer automated process without the use of any harsh processing steps, a key materials criteria when working with soft hydrogel materials. Although we have utilized model synthetic PEGDA and naturally-derived GelMA hydrogels in this work, we anticipate that HLP can be extended to other photosensitive materials.

6.3. HLP can fabricate hollow micro-features at any depth within a 3D structure

Laser based methods such as MPA depend heavily on the optical properties (transparency, absorption, scattering) of the material. ^[15–17] Low laser penetration depths limits the processing range of subtractive ablation within the materials. In contrast, HLP technology allows the fabrication of hollow micro-features at any depth within a complex user-defined 3D microstructures. This unique feature is demonstrated by the printing of (i) an embedded hollow cube-frame deep within a Mayan Pyramid 3D structure (Figure 3C), (ii) an embedded out-of-plane microchannel within a two-well PEGDA chip (Figure 4B), and (iii) 3D logpile structures at different depth/heights (Figure 6). The sequential additive-subtractive modes make this technique more-or-less independent of stringent processing depth limits. HLP can be potentially used to fabricate 3D structures using less transparent materials as well, thus decreasing the dependence of laser penetration depths on the optical properties of the materials.

6.4. HLP as compared to current light based multi-scale fabrication methods

Laser based methods that combine two separate additive processing steps or methods have been used to fabricate multiscale 3D structures.^[18,19–21] For instance, SLA was used to print multiscale surface features with a resolution of 37 μm by adaptively switching the laser spot and slice layer thicknesses.^[19] Shaped laser beams with adaptive layer thicknesses were used to print 3D multiscale structures with a resolution of 30 μm, although hollow microscale features were not reported.^[20] Large area multiscale printing was demonstrated by synchronizing linear scanner with high speed capability of galvano scanners.^[21] As compared to the methods described above, HLP can print 3D multiscale structures with a smallest feature size of 3 μm. Figure 6 highlights the quick fabrication of 3D multiscale structures with superior design flexibility.

6.5. HLP as compared to current light based multi-material fabrication methods

Few light-based methods have been adopted for multi-material printing as explained below.^[22–27] SLA and DMD based optical lithography have been used to print multiple low viscosity resins,^[23] and hydrogel-based materials.^[24] To improve the fabrication speed, DMD-SLA was recently combined with a microfluidic device^[25] and an air-jet^[26] to achieve automated and quick material exchanges. Recently, a commercial direct laser writing system was combined with a microfluidic chamber to enable 3D multimaterial printing.^[27] As compared to the current methods, HLP enables multi-material printing in additive-additive (CLIP-CLIP), additive-additive (CLIP-MPP), and additive-subtractive (CLIP-MPA) modes (Figure 8). By combining CLIP with MPP and MPA, HLP achieves a suitable trade-off between the use of multi-materials, overall throughput, size-range and feature resolution, factors critically important in manufacturing 3D multiscale structures.

The HLP technology can be further improved to enable new capabilities. For instance, the maximum size of printable 3D structure can be increased by simply modifying the polymer chamber, and by adding new inlet channels can be used to increase the number of materials. Additionally, the speed of multi-material HLP can be further improved by adopting a continuous stop-flow lithography method and automation of fluid exchange and printing processes.^[28] Furthermore, future studies can be made to better the understand the ablation mechanism, which is not currently known. Based on known literature, the formation of shockwaves and cavitation, likely led to a disruptive breakdown of the hydrogel matrix and resulted in the generation of voids during the subtractive ablation mode of HLP, however systematic studies are required.^[11] In the future, it is also advisable to study the chemical composition of the ablated regions, which is at this point difficult due to their microscale size of the ablation voids and mechanically weak nature of hydrogel materials.

In essence, we anticipate that HLP has the potential to revolutionize the ability to make 3D multiscale multi-material structures, specifically those structures that consist of internal or embedded hollow features that cannot be made using current technologies.

7. Conclusion

Due to material incompatibilities and significant differences in laser processing requirements of additive and subtractive processes, only few research groups have investigated hybrid fabrication approaches. In this work, we have designed and developed a new HLP technology that seamlessly combines additive crosslinking and subtractive ablation modes of femtosecond laser to achieve the printing of 3D multiscale multi-material structures using difficult-to-process hydrogel materials. Quick fabrication of multiscale structures with embedded hollow microfeatures demonstrates superior design flexibility of HLP as compared to conventional lithography methods, with a resolution close to that achieved by lithography. The ability to print multi-materials in additive-additive and additive-subtractive modes demonstrates the fabrication versatility of HLP. This capability can be potentially used to print 3D multi-material hydrogel-based structures for a variety of applications in biomedical sciences, microfluidics, soft robotics, optics, photonics and other application areas.

8. Experimental Section

Optical setup for Hybrid Laser Printer (HLP).

The schematic of the custom-built hybrid laser printing setup is shown in Figure 1. Both additive and subtractive modes of HLP utilize a femtosecond (fs) laser source (Coherent, Chameleon-Ultra Ultrafast Ti:Sapphire) capable of producing 150-fs wide pulses at a repetition rate of 80 MHz, with tunable fundamental wavelength ranging from 690–1040 nm. In the additive mode of HLP (shown by blue arrow in Figure 1), fs-laser beam is passed through a second harmonic generator (SHG) (Autotracker, Coherent) to obtain an ultraviolet/near visible wavelength. A shutter (SH05, Thorlabs Inc) is placed after SHG generator and 2f-transfer lens (f=40 mm and 200 mm) assembly, which collimates and expands the laser beam. A pin hole of 25 μm is used to spatially clean the laser beam. The Gaussian intensity distribution of the laser beam is changed to hat-top intensity distribution using an engineered diffuser (RPC photonics Inc.) in a rotating mount and the beam is spatially modulated via a DMD (0.95” 1080p UV DMD, DLi Inc). DMD is a spatial light modulator consisting of an array of micromirrors. User can selectively switch mirrors into either an ON state or an OFF state to modulate laser beam in user-defined light-patterns using a computer generated digital mask. A DMD-modulated fs-laser beam is projected to the sample with the infinity corrected projection optics. Laser beam modulated by the DMD can be used to crosslink photosensitive hydrogels in both continuous and layer-by-layer manner, with a feature size ranging from ~30 μm to 1 cm. A custom written MATLAB code is used to slice a 3D model of the structure and to obtain a series of DMD mask.

The subtractive mode of HLP utilizes the fs direct laser writing to fabricate structure using multiphoton ablation (MPA) process. However, this mode can also be used as multiphoton polymerization (MPP) mode simply by adjusting the laser intensity. The fs laser beam tuned at 800nm is passed through an assembly of a half-wave plate and a polarizing beam splitter, which is used to adjust the intensity of the incident laser beam. The laser beam is spatially cleaned and collimated using a 2f-transfer lens (f=40 mm and 200 mm) assembly and a 25 μm pinhole. The beam is directed towards the dichroic filters (Semrock, FF395/495/610-Di01–25×36) which reflects the laser beam towards objective lens (Carl Zeiss, 50X, 0.55 NA). The objective lens focuses the laser beam into the previously crosslinked hydrogel layer. Material from predefined region is ablated when the laser intensity exceeds the ablation threshold of the crosslinked hydrogel layer. The in situ fabrication process can be monitored in real-time using a microscopy arm incorporated in the setup. A custom-written LabVIEW program is used to synchronize the additive and subtractive processes, including switching the DMD masks, turning the shutter ON and OFF, and moving of XYZ stages. It is worth noting, that for all the experiments, laser power was measured at the back aperture of the objective lens, however for the additive crosslinking, power (P_add) was measured just before the DMD. The coverslips used in this work were methacrylated to ensure adhesion of crosslinked structure.

Meaurement of linewidth.

The linewidth of the structure was imaged by Zeiss microscopy and measured via ImageJ software. Line profiles of the structures were plotted and full width at half maxima (FWHM) was recorded to obtain the linewidth. The linewidth of the structures was averaged from 5 line profiles.

Surface modification of glass substrate.

Glass coverslips (18 mm × 18 mm, No.1; Globe Scientific) were surface modified with methacrylated function group to ensure adhesion of crosslinked structure to the glass during HLP printing. Briefly, glass coverslips were treated in piranha solution (sulfuric acid and hydrogen peroxide, 7:3 ratio) for 20 minutes, washed with MilliPore water and 100% ethanol (Fisher Scientific, Pittsburgh, PA), blow-dried using compressed air, and soaked in a bath containing 85 mM 3-(trimethoxysilyl) propyl methacrylate (Fluka, St. Louis, MO) in ethanol with acetic acid (pH 4.5) for 12 hours at room temperature, and subsequently washed in 100% ethanol and air dried.

LAP photoinitiator synthesis.

Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) was synthesized in a two-step process based on an established method. ^[29] At room temperature and under argon, 2,4,6-trimethylbenzoyl chloride (4.5 g, 25 mmol) was added dropwise to continuously stirred dimethyl phenylphosphonite (4.2 g, 25 mmol). This solution was stirred for 24 hours before adding an excess of lithium bromide (2.4 g, 28 mmol) in 50 mL of 2-butanone to the reaction mixture at 50°C, to obtain a solid precipitate after 10 minutes. The mixture was then cooled to room temperature, allowed to rest overnight and then filtered. The filtrate was washed with 2-butanone (3 × 25 mL) to remove unreacted lithium bromide and dried under vacuum to give LAP (6.2 g, 22 mmol, 88% yield) as a white solid.

Preparation of PEGDA and GelMA prepolymer solutions.

Different concentrations of Poly (ethylene glycol) diacrylate (PEGDA, Mw=700, Sigma-Aldrich) prepolymer solutions were prepared for this project. First, LAP (0.4 gm) was added to 4 ml of Millipore water and mixed thoroughly using Vortex mixer (Vortex-Genie 2 from Fisher Scientific). Second, 36 ml of 100% PEGDA hydrogel is added to the 4 ml of LAP solution to prepare 90% PEGDA solution and further mixed in Vortex mixer, and incubated at 37 °C to remove the bubbles. Similarly, 10% and 50% PEGDA prepolymer solutions were also prepared.

Gelatin Methacrylate (GelMA) macromer was synthesized as described previously.^[30] Briefly, 200 ml Dulbecco’s phosphate buffered saline (DPBS, Gibco) was heated to 40 ℃, and was used to dissolve 10 gm of porcine skin gelatin (Sigma Aldrich, St. Louis, MO). Further, methacrylic anhydride was subsequently added to the solution and stirred for 3 hours. The unreacted groups from the solution was removed by dialyzing the mixture against distilled water for a week at 40 ℃. Next, the dialyzed GelMA was lyophilized in a freeze dryer (Labconco, Kansas City, MO) for another week. A stock solution was prepared by mixing 0.75 g freeze dried GelMA with 10 ml of deionized water at 40 ℃ and 0.01875 gm of LAP photoinitiator to obtain a 10% (w/v) GelMA with 0.25% LAP (w/v).

Cell culture.

All the cells used in the experiment were cultured under standard cell culture conditions (5% CO₂; 37 °C). MLO-Y4 osteocytes were maintained in α-Modified Eagle Medium (α-MEM, Gibco) supplemented with 2.5% heat inactivated FBS, 2.5% calf serum (CS), 1% P/S and 1% Glutamax on rat tail type I collagen-coated flasks, and passage 9–15 were used. Saos-2 human osteosarcoma cells were maintained in DMEM supplemented with 10% FBS, 1% P/S and 1% Glutamax, and were passaged when 70–80% confluence was reached.

Cell culture within PEGDA chips.

Multi-wells PEGDA chips were designed and printed using HLP. The PEGDA multi-well chips were coated with rat tail type I collagen at 37 °C overnight and then flushed with DPBS before seeding either MLO-Y4 osteocytes or Saos-2 osteosarcoma cells at a final concentration of 1 M/ml. Specifically, a 2 μl cell solution was added to the chips, and placed in the incubator (37 °C) for 20 minutes to allow for sufficient attachment of the cells, before adding medium to the entire chip and culturing them under standard culture conditions. Live/Dead assay kit (Invitrogen) (Calcein AM, Ethidium homodimer) was used to quantify the viability of cells after 3 days of culture. Briefly, a solution of 0.5 μl/ml calcein AM and 2 μl/ml Ethidium homodimer in BME was pipetted into PEGDA biochips, incubated at 37 °C for 45 minutes before capturing images using an inverted microscope (Nikon Eclipse). To determine cellular morphology, cells cultured within chips were fixed with 4% formaldehyde (Invitrogen, Carlsbad, CA) for 30 mins, soaked in 2% Triton X-100 in DPBS for 30 minutes, and subsequently stained with phalloidin (Alexa-Fluor 568, Invitrogen) (1/100 dilution in DPBS) (45 mins at RT) and with 1.25 μg/ml DAPI (Life Technologies) (5 minutes at RT) to visualize f-actin and cell nuclei respectively.

Digital optical microscope imaging.

The fabricated structures were imaged and characterized using a digital optical microscope (HIROX, KH-8700). The MXG-2016Z /MX-2016Z 20–160x objective lens was used to image the fabricated structures, which provided a resolution of 1.35 μm.

Figure 7.

(A) Syringe pump system for multiscale, multimaterial fabrication. Three different inlets and three outlets were used for exchange of printing material. Nitrogen/water inlets and two water outlets were meant for washing and drying steps. (B) Side view of the fluid chamber, which depicts 5 inlets/5 outlets and fabrication area. Cross-section view (at marked green dotted line) shows the sloped floor design that facilitates easy exchange of prepolymer solutions during printing.